Merge pull request #40 from id774/master

Add Filter::Sanitize

Gemfile

@@ -12,6 +12,7 @@ gem 'gcalapi'
 gem 'xml-simple'
 gem 'feedbag'
 gem 'nokogiri'
+gem 'sanitize'
 gem 'twitter'
 gem 'weather_hacker'
 gem 'pocket-ruby'

doc/PLUGINS

@@ -281,6 +281,22 @@ FilterRand
   - module: FilterRand
 
 
+FilterSanitize
+--------------
+[Path]
+ /plugins/filter/sanitize.rb
+
+[Abstract]
+ Strip the html tags.
+ Reference: https://github.com/rgrove/sanitize
+
+[Syntax]
+  - module: FilterSanitize
+    config:
+      mode: basic or relaxed or restricted
+            (default: restricted)
+
+
 FilterFullFeed
 --------------
 [Path]

doc/PLUGINS.ja

@@ -282,6 +282,22 @@ FilterRand
   - module: FilterRand
 
 
+FilterSanitize
+--------------
+[パス]
+ /plugins/filter/sanitize.rb
+
+[概要]
+ HTML タグを除去する
+ 参照 https://github.com/rgrove/sanitize
+
+[レシピ記法]
+  - module: FilterSanitize
+    config:
+      mode: basic or relaxed or restricted
+            (デフォルト restricted)
+
+
 FilterFullFeed
 --------------
 [パス]

plugins/filter/sanitize.rb

@@ -0,0 +1,41 @@
+# -*- coding: utf-8 -*-
+# Name::      Automatic::Plugin::Filter::Sanitize
+# Author::    774 <http://id774.net>
+# Created::   Jun 20, 2013
+# Updated::   Jun 20, 2013
+# Copyright:: 774 Copyright (c) 2013
+# License::   Licensed under the GNU GENERAL PUBLIC LICENSE, Version 3.0.
+
+module Automatic::Plugin
+  class FilterSanitize
+    require 'sanitize'
+
+    def initialize(config, pipeline=[])
+      @config = config
+      @pipeline = pipeline
+      case @config['mode']
+        when "basic"
+          @mode = Sanitize::Config::BASIC
+        when "relaxed"
+          @mode = Sanitize::Config::RELAXED
+        else
+          @mode = Sanitize::Config::RESTRICTED
+      end
+    end
+
+    def run
+      @return_feeds = []
+      @pipeline.each {|feeds|
+        return_feed_items = []
+        unless feeds.nil?
+          feeds.items.each {|feed|
+            feed.description = Sanitize.clean(feed.description, @mode) unless feed.description.nil?
+          }
+          @return_feeds << feeds
+        end
+      }
+      @return_feeds
+    end
+
+  end
+end

spec/plugins/filter/sanitize_spec.rb

@@ -0,0 +1,128 @@
+# -*- coding: utf-8 -*-
+# Name::      Automatic::Plugin::Filter::Sanitize
+# Author::    774 <http://id774.net>
+# Created::   Jun 20, 2013
+# Updated::   Jun 20, 2013
+# Copyright:: 774 Copyright (c) 2013
+# License::   Licensed under the GNU GENERAL PUBLIC LICENSE, Version 3.0.
+
+require File.expand_path(File.dirname(__FILE__) + '../../../spec_helper')
+
+require 'filter/sanitize'
+
+describe Automatic::Plugin::FilterSanitize do
+  context "It should be sanitized" do
+    subject {
+      Automatic::Plugin::FilterSanitize.new(
+        {},
+        AutomaticSpec.generate_pipeline {
+          feed {
+            item "http://testsite.org", "hoge",
+            "<a>fuga</a>",
+            "Mon, 07 Mar 2011 15:54:11 +0900"
+          }
+        }
+      )
+    }
+
+    describe "#run" do
+      its(:run) { should have(1).feeds }
+
+      specify {
+        subject.run
+        subject.instance_variable_get(:@return_feeds)[0].items.
+          count.should == 1
+        subject.instance_variable_get(:@return_feeds)[0].items[0].description.
+          should == 'fuga'
+      }
+    end
+  end
+
+  context "It should not be sanitized in basic mode" do
+    subject {
+      Automatic::Plugin::FilterSanitize.new(
+        {
+          'mode' => "basic"
+        },
+        AutomaticSpec.generate_pipeline {
+          feed {
+            item "http://testsite.org", "hoge",
+            "<a>fuga</a>",
+            "Mon, 07 Mar 2011 15:54:11 +0900"
+          }
+        }
+      )
+    }
+
+    describe "#run" do
+      its(:run) { should have(1).feeds }
+
+      specify {
+        subject.run
+        subject.instance_variable_get(:@return_feeds)[0].items.
+          count.should == 1
+        subject.instance_variable_get(:@return_feeds)[0].items[0].description.
+          should == '<a rel="nofollow">fuga</a>'
+      }
+    end
+  end
+
+  context "It should not be sanitized in restricted mode" do
+    subject {
+      Automatic::Plugin::FilterSanitize.new(
+        {
+          'mode' => "restricted"
+        },
+        AutomaticSpec.generate_pipeline {
+          feed {
+            item "http://testsite.org", "hoge",
+            "<a>fuga</a>",
+            "Mon, 07 Mar 2011 15:54:11 +0900"
+          }
+        }
+      )
+    }
+
+    describe "#run" do
+      its(:run) { should have(1).feeds }
+
+      specify {
+        subject.run
+        subject.instance_variable_get(:@return_feeds)[0].items.
+          count.should == 1
+        subject.instance_variable_get(:@return_feeds)[0].items[0].description.
+          should == 'fuga'
+      }
+    end
+  end
+
+  context "It should not be sanitized in relaxed mode" do
+    subject {
+      Automatic::Plugin::FilterSanitize.new(
+        {
+          'mode' => "relaxed"
+        },
+        AutomaticSpec.generate_pipeline {
+          feed {
+            item "http://testsite.org", "hoge",
+            "<a>fuga</a>",
+            "Mon, 07 Mar 2011 15:54:11 +0900"
+          }
+        }
+      )
+    }
+
+    describe "#run" do
+      its(:run) { should have(1).feeds }
+
+      specify {
+        subject.run
+        subject.instance_variable_get(:@return_feeds)[0].items.
+          count.should == 1
+        subject.instance_variable_get(:@return_feeds)[0].items[0].description.
+          should == '<a>fuga</a>'
+      }
+    end
+  end
+
+end

test/integration/test_sanitize.yml

@@ -0,0 +1,22 @@
+global:
+  timezone: Asia/Tokyo
+  cache:
+    base: /tmp
+  log:
+    level: info
+
+plugins:
+  - module: SubscriptionFeed
+    config:
+      feeds:
+        - http://blog.id774.net/post/feed/
+
+  - module: FilterSanitize
+    config:
+      mode: restricted
+
+  - module: StoreFullText
+    config:
+      db: test_sanitize.db
+
+  # - module: PublishConsole